Adiponectin Promotes Neurogenesis After Transient Cerebral Ischemia Through STAT3 Mediated BDNF Upregulation in Astrocytes

Newborn neurons from the subventricular zone (SVZ) are essential to functional recovery following ischemic stroke. However, the number of newly generated neurons after stroke is far from enough to support a potent recovery. Adiponectin could increase neurogenesis in the dentate gyrus of hippocampus in neurodegenerative diseases. However, the effect of adiponectin on the neurogenesis from SVZ and the functional recovery after ischemic stroke was unknown, and the underlying mechanism was not specified either. The middle cerebral artery occlusion model of mice was adopted and adiponectin was administrated once a day from day 3 to 7 of reperfusion. The levels of BDNF and p-STAT3 were detected by western blotting on day 7 of reperfusion. The virus-encoded BDNF shRNA with GFAP promoter and a STAT3 inhibitor Stattic were used, respectively. Neurogenesis was evidenced by the expression of doublecortin and 5-bromo-2ʹ-deoxyuridine (BrdU) labelling and brain atrophy was revealed by Nissl staining on day 28 of reperfusion. Neurological functional recovery was assessed by the adhesive removal test and the forepaw grip strength. We found that adiponectin increased both the doublecortin-positive cells and NeuN/BrdU double-positive cells around the injured area on day 28 of reperfusion, along with the improved long-term neurological recovery. Mechanistically, adiponectin increased the protein levels of p-STAT3 and BDNF in astrocytes on day 7 of reperfusion, while silencing BDNF diminished the adiponectin-induced neurogenesis and functional recovery. Moreover, inhibition of STAT3 not only prevented the increase of BDNF but also the improved neurogenesis and functional recovery after stroke. In conclusion, adiponectin enhances neurogenesis and functional recovery after ischemic stroke via STAT3/BDNF pathway in astrocytes.


Introduction
Stroke is a leading cause of adult disabilities worldwide. As the application of recanalization therapies is very limited due to the strict screening criteria [1] and most neuroprotective strategies for stroke failed, the spontaneous neural repair Liang Yu, Jiajia Wang and Ying Xia equally contributed to this work.

3
after cerebral ischemia may provide enlightenment for stroke research. As one of the important processes of spontaneous neural repair, endogenous neurogenesis in the subventricular zone (SVZ) is stimulated and contributes to neurological recovery after cerebral ischemia [2]. But the advantages of such endogenous neurogenesis are not exploited to the full for stroke recovery, as proximately 80% of the newborn neurons induced by stroke die within 2 weeks, only very few survive and integrate into functional neural network [3]. Therefore, approaches to promote the neurogenesis and the survival of newly generated neurons from neural stem cells are of significance in improving functional recovery after stroke. Adiponectin (ADPN), an adipocyte-derived secretory protein with a variety of physiological roles in nervous systems, could interestingly boost adult neurogenesis in the dentate gyrus of hippocampus in neurological diseases [4]. This enlightened us that ADPN may promote neurogenesis in SVZ and neurological recovery after stroke.
Adult neurogenesis is strictly regulated by microenvironment cues in the neurogenesis niches. One of the important factors is brain-derived neurotrophic factor (BDNF), which has a pivotal role in neuronal generation and maturation, synapse formation and synaptic plasticity in central nervous system, is indispensable for neural repair after cerebral injury. Astrocytes are the main source of BDNF and take a great part in maintaining the microenvironmental homeostasis and facilitating neurogenesis in SVZ [5]. Astrocytes are reactivated after nervous injury and the neural reparative phenotype is predominantly under the regulation of signal transducer and activator of transcription 3 (STAT3) signaling which serves as an important upstream regulator of BDNF [6,7]. However, whether the STAT3/BDNF pathway was involved in ADPN-promoted neurogenesis after ischemic stroke was unknown.
In the present study, we tried to reveal whether ADPN treatment could promote neurogenesis after experimental stroke, as well as the potential role of STAT3/BDNF pathway in astrocytes after stroke.

Animals
Male C57BL/6 mice, 8-10 weeks old, weighing from 20 to 22 g, without prior experimental use, were provided by the Experimental Animal Center of Air Force Military University (former The Fourth Military Medical University). A total of 120 mice were used, and mice were kept in a standardized housing condition with an environmental temperature of 22 ± 2 ºC and a relative humidity of 50 ± 1%. The animal procedures were performed strictly consistent with the regulations and guidelines of Air Force Military University institutional animal care and according to the AAALAC and the Institutional Animal Care and Use Committee guidelines.

Experimental Design
In the first part, in order to verify the effect of exogenous ADPN on the neurogenesis and functional outcomes after ischemic stroke, animals were allocated into three groups with a random fashion: Sham, Vehicle and ADPN groups. In the ADPN group, ADPN (HY-P7358, MedChemExpress, California, USA) intravenous administration with a dose of 100 μg/kg was started on the third day after MCAO, once a day for 5 consecutive days. The dose was according to a previous study [8]. In vehicle group, 5 ml/kg saline was intravenously administrated. Neurogenesis by doublecortin (DCX) expression and 5-bromo-2ʹ-deoxyuridine (BrdU) staining, and brain atrophy were analyzed 28 days after MCAO. Neurological performance was recorded on day 7, 14 and 28 after reperfusion. In addition, the mRNA and protein levels of BDNF, protein levels of STAT3 and phosphorylated STAT3 at Tyr-705 and Ser-727 were investigated in area adjacent to SVZ on day 7 after MCAO.
In the second part, to elucidate the role of STAT3/BDNF signaling in ADPN-promoted neurogenesis, we injected Stattic into lateral cerebral ventricle to inhibit STAT3. In addition, the recombinant adeno-associated virus (AAV) 9 encoding BDNF-shRNA with GFAP promoter was also used. Animals were randomly divided into four groups: ADPN + DMSO, ADPN + Stattic, ADPN + control virus and ADPN + BDNF shRNA. All animals were subjected to MCAO and received ADPN treatment as forementioned. In ADPN + Stattic group, Stattic was injected into right lateral cerebral ventricle at 30 min before ADPN administration, and saline containing 50% of DMSO was used as the vehicle control of Stattic; in ADPN + BDNF shRNA group, the virus was injected into the area adjacent to SVZ 3 weeks before MCAO, and the control virus containing the scramble RNA was used as vector control. Brain atrophy and neurogenesis were analyzed 28 days after MCAO. Each protein level was detected at the forementioned timepoints by western blotting and immunofluorescence staining. In addition, neurological performance was recorded on day 7, 14 and 28 of reperfusion.

Transient Cerebral Ischemia and Reperfusion
Mice were given free access to food and tap water before procedure. The transient focal cerebral ischemia injury was performed according to the intraluminal monofilament occlusion of middle cerebral artery (MCAO) as we previously reported [9]. Briefly, animals were anesthetized with 1.5% isoflurane (20181501, RWD Life Science, Shenzhen, China). A commercial suture for MCAO (RWD Life Science) was used to abrupt the blood flow to the right MCA. The blood flow to the right MCA was blocked for 1 h and the suture was then removed to restore the blood flow. The body temperature of mice was monitored by a rectal probe and maintained to 37 ± 0.5 ºC by a heating blanket. A laser Doppler sensor (periFlux System 5000, PERIMED, Stockholm, Sweden) for regional cerebral blood flow monitoring was placed on the skull surface of mice, 2 mm caudal and 4 mm lateral to the right of bregma. Mice with 80% decrease and 70% recovery of the blood flow were qualified to the subsequent experiment as the ischemic stroke mice.

Drug Administration
ADPN was prepared in 0.9% saline to a concentration of 25 μg/ml. Intravenous ADPN treatment was started on the third day after MCAO, once per day for 5 consecutive days through tail vein. The dose of ADPN was chosen based on the previous study which screened the doses of ADPN on the ischemic cerebral injury model and proved that the optimal dose was 100 μg/kg [8]. A STAT3 inhibitor Stattic (HY-13818, MedChemExpress), was dissolved in saline containing 50% DMSO at a concentration of 30 μM and was microinjected though an implanted cannula into the right lateral cerebral ventricle with a volume of 2 μl. Stattic treatment was according to a previous study [10]. The cannula was connected with a polyethylene tube to a Hamilton syringe. The injection speed was 1 μl no less than 120 s. The inhibitory effect of Stattic was shown in Supplementary Fig. S1.

Stereotaxic Surgery
After anesthesia, mouse head was fixed on a stereotaxic frame (Stoelting, Kiel, Wisconsin, USA). A midline incision was made, then the skin and underlying periosteum were retracted. The cannula was placed into the right lateral cerebral ventricle according to the stereotaxic coordinates: 0.2 mm anterior to the bregma, 1.0 mm lateral from midline, and 2.0 mm in depth from skull surface. The cannula was anchored to the surrounding skull with dental cement. After stereotaxic operation, animals were then placed back into the feeding cages.

Assessment of Neurological Deficits
Sensorimotor functions were evaluated with the adhesive removal test and forepaw grip strength test. Mice were trained for 3 consecutive days before MCAO to generate baseline data. Each mouse underwent both tests on day 7, 14 and 28 after MCAO.
1. The adhesive removal test was performed as previously described [11]. In brief, adhesive tapes (0.3 × 0.4 cm) were applied to hairless part of both forepalms with equal pressure. The times to contact and removal of the tapes were recorded. Each mouse was tested three times with 120 s per trials. The data were presented as the mean time to contact and the mean time to removal the tape from three trial. 2. The forepaw grip strength test was used to assess the forepaw motor function of the mice as previously described [12]. In brief, maximum grip strength of both forepaws was recorded using a digital gauge (BIO-GS3, Bioseb, France). The mice were held by tail above a T-shaped bar connected to the force gauge. The mouse was gently lifted in a position that allowed it to reflexively grasp the bar with its both forepaws. The researcher then gently pulled the animals horizontal away from the bar within 3 s and the maximum force before losing its grasp would be recorded. The trial was repeated for three times and a few second breaks between trials were allowed to regain forepaw strength of mice.

Measurement of Atrophy Volume
The mice were deep anesthetized. The brains were removed and post-fixed with 4% paraformaldehyde for 12 h. After gradient dehydration in 20% and 30% sucrose solution, the brains were cryostat sectioned into 15-μm-thick slices serially at 200-μm distance and stained with cresyl violet (Beyotime, China). The stained samples were photographed using microscope and measured in a blind manner with image analysis software. The brain atrophy was calculated following the equation: atrophy ratio = (the contralateral hemisphere area -the ipsilateral hemisphere area)/contralateral area × 100%.

5-Bromo-2ʹ-Deoxyuridine (BrdU) Labeling and Staining
BrdU (Sigma Aldrich, 10 mg/ml in saline, and 50 mg/kg) was injected intraperitoneally from day 3 to 14 after the onset of stroke. At the end of day 28, the brains were removed and post-fixed in 4% paraformaldehyde, then dehydrated in sucrose. The brains were sectioned into 12-μm-thick slices serially by a cryostat microtome (CM1950, Leica Biosystem, Frankfurt, Germany

Immunofluorescence Staining
Immunofluorescence staining was performed on frozen coronal cerebral sections of mice. Mice were fixed with 4% paraformaldehyde after deep anesthesia. After postfixation for 6-8 h and dehydration in sucrose, the brains were then cut into 12-μm-thick sections. The brain sections were washed with PBS and then incubated with the following primary antibodies overnight at 4 ºC in a humidified atmosphere: rabbit anti-BDNF (

qRT-PCR
Total RNA was extracted on day 7 after ischemia from periinfarct tissue of striatum with trizol reagent (

Analysis of Newborn Neurons and DCX Positive Cells
Three frozen slices from each mouse were used for DCX or BrdU and NeuN staining. Section 1 was 1.0 mm anterior to bregma, section 2 was 0.8 mm anterior to bregma, and section 3 was 0.6 mm anterior to bregma. Three fields between lateral cerebral ventricle and the injured striatum were captured by a 20 × objective lens. The number of DCX-positive cells was obtained from the average of DCX-positive cells in 9 fields of each mouse, and each group contained 4 mouse samples. The newborn neurons in each mouse were the average number of NeuN/BrdU double positive cells in 9 fields.

Transfection of AAV
The AAV encoding BDNF-shRNA-mCherry with the promoter of GFAP was purchased from BrainVTA Co., Ltd.
(Wuhan, China). The BDNF target sequence was: 5ʹ-TAT GTA CAC TGA CCA TTA A-3ʹ. The control virus containing the scrambled RNA was also from BrainVTA. Transfection was performed by stereotactic injection (volume, 300 nl; titer, 1 × 10 12 ). The microinjection coordinate was AP 0.8 mm, ML 1.5 mm, DV 2.5 mm from bregma. The reliability of AAV was validated by immunofluorescence labeling and western blotting 3 weeks after infection, as shown in Fig. S2.

Statistical Analysis
Statistical analyses were performed using SPSS (version 19.0, IBC Corp, Armonk, NY, USA). All data were presented as mean with standard deviation. The sample size was 8 per group for neurological test and brain atrophy, and 4 per group for morphology and molecular biology. Multiple comparisons were conducted with one-way analysis of variance (ANOVA), followed by Tukey's post hoc test. And repeated measures were analyzed with two-way ANOVA followed by Bonferroni's multiple comparison for each timepoint. P values less than 0.05 were considered statistically different.

ADPN Treatment Increased Neurogenesis After Cerebral Ischemia
Neurogenesis takes a great part in spontaneous neural repair after cerebral ischemia. In order to determine the effect of ADPN on neurogenesis, we counted the BrdU-positive neurons and DCX-positive cells, and we also evaluated DCX expression in peri-infarct tissue of striatum on day 28 1F) compared with those in Vehicle groups. In addition, the migration of immature neurons from the SVZ to the striatum were also detected on 28 days after MCAO. As shown in Fig. 1D Fig. 2D.

ADPN Up-Regulated BDNF Expression in Astrocytes After Ischemic Stroke
We proved that ADPN promoted neurogenesis after MCAO, then we tried to unravel the role of BDNF in this process. We first evaluated the expression of BDNF in ipsilateral area adjacent to SVZ on day 7 after MCAO. We found that Vehicle group had higher expression of BDNF in both protein level ( Fig. 3C). In addition, we also revealed the cell specific expression of BDNF by immunofluorescence. The results showed that BDNF was mainly expressed on astrocytes after MCAO regardless of ADPN treatment (85.32 ± 7.64% vs 78.35 ± 9.34%, ADPN vs Vehicle, P > 0.05, Fig. 3D, E). size, which is a mirror image of the contralateral side. Time to contact, time to removal and grip strength were presented as mean with standard deviation and analyzed by two-way ANOVA followed by Bonferroni's multiple comparison for each timepoint, cerebral atrophy was analyzed by one way ANOVA followed by Tukey's post hoc test. The contact and removal times, and grip strength were the average of 3 repetitions each mouse. *P < 0.05 vs. Sham group, #P < 0.05 vs. Vehicle group, n = 8

ADPN Activated STAT3 Signaling in Astrocytes After Ischemic Stroke
As STAT3 is a key transcriptional factor of astrocytes, we further verified the effect of ADPN on STAT3 in astrocytes.  Fig. 4G, H) were further increased in astrocytes after ADPN treatment compared with the Vehicle group. These data suggested that STAT3 signaling in astrocyte was promoted on day 7 after stroke by ADPN treatment.

Inhibition of STAT3 Prevented the Up-Regulation of BDNF Induced by ADPN
In order to testify the role of STAT3 in ADPN-induced up-regulation of BDNF, we analyzed BDNF expression after pharmacological inhibition of STAT3. An inhibitor of STAT3 Stattic which inhibits the phosphorylation of STAT3 at both sites of Tyr705 and Ser727, was injected into lateral cerebral ventricle at 30 min before ADPN treatment. The expression of BDNF was evaluated on day 7 after MCAO. We found that Stattic decreased the expression of BDNF in both the protein level (0.56 ± 0.17 vs 1.26 ± 0.21, ADPN + stattic vs ADPN + DMSO, P < 0.05, Fig. 5B and C) and the mRNA level (51.87 ± 10.91% vs 125.75 ± 21.10%, ADPN + stattic vs ADPN + DMSO, P < 0.05, Fig. 5D. These data indicated that STAT3 might be the mediator of ADPN up-regulating BDNF.

Inhibition of STAT3/BDNF Pathway Prevented the Increase of Neurogenesis Induced by ADPN
In order to determine the role of STAT3/BDNF pathway in ADPN-induced neurogenesis, the STAT3-inhibitor Stattic and AAV-encoded BDNF shRNA were used, respectively. Newborn neurons were counted around the injured area in ipsilateral striatum on day 28 Fig. 6G). These data demonstrated that the STAT3/BDNF signaling was involved in ADPN-promoted neurogenesis.

Inhibition of STAT3/BDNF Signaling Pathway Prevented the Improvement of Stroke Recovery after ADPN Treatment
We then evaluated the long-term functional outcomes after MCAO in the presence of blocking STAT3/BDNF signaling by using the STAT3-inhibitor Stattic or AAV-encoded BDNF shRNA. The neurological performance and brain atrophy were evaluated. Both Stattic and BDNF shRNA prolonged both the time to contact (P < 0.05, Fig. 7B) and the time to removal (P < 0.05, Fig. 7C). Meanwhile, the increase of forelimb grip strength by ADPN was prevented by Stattic and BDNF shRNA on day 14 and 28 after stroke (P < 0.05, Fig. 7D). Moreover, Stattic and BDNF shRNA prevented the improvement of brain atrophy induced by ADPN treatment (20.92 ± 4.86 vs 13.61 ± 3.46, ADPN + Stattic vs ADPN + DMSO, P < 0.05; 18.66 ± 3.61 vs 12.54 ± 2.89, ADPN + BDNF shRNA vs ADPN + control virus, P < 0.05, Fig. 7E, F). These data demonstrated that STAT3/BDNF signaling was involved in ADPN-improved long-term functional outcomes after stroke.

Discussion
In the present study, we found that ADPN treatment from day 3 after MCAO enhanced neurogenesis and improved functional outcomes in mice. Mechanistically, ADPN treatment increased the expression of BDNF and promoted STAT3 activation in astrocytes. Additionally, inhibition of STAT3 prevented the upregulation of BDNF induced by ADPN treatment. We also found that both the inhibition of STAT3 and the silence of BDNF in astrocytes opposed the 1 3 promotion of neurogenesis and long-term functional recovery induced by ADPN treatment. These data suggest that ADPN could promote neurogenesis and functional recovery after stroke, with the involvement of STAT3/BDNF signaling, which implies a valuable method to improve the recovery of stroke.
Lower level of serum ADPN after stroke was associated with higher mortality [13]. ADPN was also negatively correlated to the damage size and functional deficits after stroke [14]. These clinical data indicate that ADPN may have an important role in stroke. Indeed, in experimental ischemic stroke, early or prophylactic application of exogenous ADPN exerted neuroprotection on mice [8,15]. Here, we focused on the neural recovery potential of ADPN. We initiated ADPN treatment on day 3 after MCAO, which was based on the time course when neurogenesis is activated after stroke [16]. By using BrdU-labelling to trace cells with occurred division and co-labelled with NeuN, we showed the matured newborn neurons in the peri-infarct area, which would replace the dead neurons after injury. We also showed the immature neurons with a common marker DCX, which was consistent with previous studies confirmed that neurogenesis was increased and very necessary to functional recovery in rodents and patients after stroke [17,18]. Due to the limited intrinsic reparative capacity of endogenous neurogenesis after stroke [19], methods to promote neurogenesis had the potential to improve the recovery of stroke, such as physical exercise and electroacupuncture [20,21]. It was also confirmed that peripheral B lymphocytes could improve functional recovery after stroke by supporting neurogenesis [22]. Likewise, in the present study, we confirmed the increase of newborn neurons in striatum and cortex simultaneously in the ischemic boundary zone after MCAO, even though there were very few newborn neurons in cortex (Fig. S3). We therefore mainly observed neurogenesis in striatum because there were quite numerous newborn neurons in striatum which was proven to be closely related to neurological function recovery [23]. In fact, the promotion of neurogenesis by ADPN have already been reported in mouse dentate gyrus, which was associated with its antidepressant effect [4], while in our study, we re-confirmed the increase of neurogenesis by ADPN in dentate gyrus (data not shown). Besides, we also found ADPN promoted neurogenesis in SVZ after stroke and associated it with improved sensorimotor function recovery. The endogenous physiological roles of ADPN forecast its minimal side effects. The main concern of ADPN is that it could hardly pass through the BBB when it is intact. However, the BBB would remain open for weeks after stroke [24]. It was already confirmed that ADPN could enter into brain through the leaky BBB and exert neuroprotection after stroke [25]. Besides, fusing with cell-penetrating peptides may overcome this drawback of ADPN.
Microenvironment in the neurogenesis niche was a determinant factor of adult neurogenesis [26]. BDNF, composed of 252 amino acids and universally expressed in the nervous system, is one of the most studied neurotrophic factors in neurogenesis. BDNF plays an important role in neurodevelopment and maintaining brain functions by regulating neural stem cells proliferation and differentiation and neural plasticity [27]. In ischemic stroke, BDNF was upregulated and contributed to enhanced neurogenesis and improved neurological functional recovery [28]. In this study, we reconfirmed the stimulated expression of BDNF after stroke. We also proved that BDNF was further increased by ADPN in SVZ on day 7 after stroke, consistent with a previous study [25]. However, the cellular source of increased BDNF after stroke was undetermined. Under physiological conditions, BDNF expression was low in SVZ and mainly on astrocytes and ependymal cells [29]. Here, we confirmed that BDNF responsible for neurogenesis in SVZ after stroke was mainly from the surrounding astrocytes. By using the astrocyte-specific AVV containing BDNF shRNA, we silenced the expression of BDNF in astrocytes and found that the enhancement of neurogenesis and the improvement of neurological function by ADPN treatment was prevented, which indicated that BDNF-triggered neurogenesis was required for ADPN treatment to improve stroke recovery. This was similar to a previous study that overexpression of BDNF in astrocytes promoted neurogenesis in hippocampus [30]. BDNF binds to two receptors: the high-affinity tropomyosin-related receptor B (TrkB) and the low-affinity pan-neurotrophin receptor (p75NTR). TrkB was expressed on proliferating cells in SVZ and enhanced the proliferation of neural stem cells in SVZ [31], while p75NTR on migrating neuroblasts and promoted migration of neuroblasts [32]. In the present study, we showed that ADPN not only promoted the proliferation of neuronal stem cells but also the migration as both the total number of DCX-positive cells and DCX-positive cells in areas distant from SVZ were markedly increased. Therefore, it could be speculated that both receptors of BDNF participated in ADPN-induced neurogenesis after stroke, which warranted future verification.
STAT3, one of STAT family member, participates in the regulation of cellular growth, differentiation, and survival. Tyr705 phosphorylation of STAT3 was primarily mediated by Janus kinase and Ser727 phosphorylation by MAPKs or ERK, which was required for STAT3 to achieve its maximal function [33]. Phosphorylation of STAT3 was significantly increased in the ischemic penumbra and peaked at 24 h after Fig. 4 The effect of ADPN treatment on the activation STAT3 in astrocytes in the peri-infarct striatum. A The representative images of western blotting of STAT3, p-STAT3(Ser705) and p-STAT3(Tyr705). B Quantification of p-STAT3(Ser705). C Quantification of p-STAT3(Tyr705). D Quantification of total STAT3. E The representative immunofluorescence images of p-STAT3(Tyr705). F Quantification of relative fluorescence intensity of p-STAT3(Tyr705) in astrocytes. G The representative immunofluorescence images of p-STAT3(Ser727). H Quantification of relative fluorescence intensity of p-STAT3(Ser727) in astrocytes. Data were presented as mean with standard deviation and analyzed by one way ANOVA followed by Tukey's post hoc test. Each experiment was repeated 3 times. *P < 0.05 vs. Sham group, #P < 0.05 vs. Vehicle group, n = 4 ◂ stroke, and then decreased gradually [34]. It is interesting that the role of STAT3 was related to the phase of stroke: while activation of STAT3 at acute phase of stroke causes tissue damage, activation of STAT3 in delayed stage promotes tissue repair such as angiogenesis and neurogenesis [35]. In this study, we found that administration of exogenous ADPN increased the level of p(Tyr-705) STAT3 and p(Ser727) STAT3 in SVZ on day 7 after stroke and inhibition of STAT3 prevented the ADPN-induced neurogenesis, consistent with the neural reparative role of STAT3 in the subacute phase of stroke. The role of STAT3 in promotion of neurogenesis after stroke was also confirmed by another study [36]. Our recent study showed that activation of STAT3 promoted the polarization of A1 astrocytes toward A2 astrocytes to attenuate the cerebral injury [37]. However, other's work suggested astrocyte-specific deletion of STAT3 exerted neuroprotection on stroke animals [38]. This inconformity of the role of STAT3 may be explained with different stages of stroke as well. As for the regulation of BDNF by STAT3, it is reasonable to speculate that STAT3 governs the BDNF expression in astrocytes during neural repair, as STAT3 is a key transcriptional factor of reparative astrocytes [39]. By using a STAT3 inhibitor Stattic, we found ADPN-increased BDNF in astrocytes was prevented, confirming BDNF as a downstream molecule of STAT3 in ADPN treatment, also consistent with the previous study [7].
Further question remains that the exact relationship between ADPN and STAT3 is unclear. Janus kinase 1 was the direct activator of STAT3 [40]. In this study, we tested the activation of JAK-1 and found that JAK-1 was activated after MCAO, and adiponectin could further promote its activation after MCAO (Fig. S4). This suggested that the canonical JAK/STAT pathway was stimulated by administration of ADPN. ADPN has various roles dependent on its specific receptors: ADIPOR1, ADIPOR2 and T-cadherin. ADIPOR1 and ADIPOR2 both are membrane proteins and expressed abundantly in brain [41]. We previously confirmed that activation of ADIPOR1 was neuroprotection against experimental stroke [42]. Recent studies showed that ADPN could promote neurogenesis in dentate gyrus via ADIPOR1 [43]. In the present study, we observed that adiponectin receptor 1 was increased after   The effect STAT3 inhibition or BDNF silencing on the longterm functional outcomes after ADPN treatment. A Schematic experiment design for B-F. B Quantification of time to contact measured on day 7, 14 and 28 d of reperfusion. C Quantification of the time to removal after MCAO. D Quantification of forelimb grip strength after MCAO. E The brain atrophy on day 28 of reperfusion. F Representative images of brain atrophy in different groups. Neurological tests were presented as mean with standard deviation and analyzed by two-way ANOVA followed by Bonferroni's multiple comparison. Cerebral atrophy was analyzed by one way ANOVA followed by Tukey's post hoc test. The contact and removal times, and grip strength were the average of 3 repetitions each mouse. *P < 0.05 vs. ADPN + DMSO, # P < 0.05 vs ADPN + control virus, n = 8 MCAO, and increased further when treated after adiponectin (S5). It was reported that ADIPOR1 could activate the JAK1/STAT3 pathway in hepatocellular carcinoma [44]. These results imply that ADPN may also activate JAK1/ STAT3 pathway through ADIPOR1 after ischemic stroke. Nevertheless, the roles of the adaptor protein APPL1 that directly interacted with ADIPOR1 [45] and its primary downstream kinase AMPK [46] in activating STAT3 should be checked further.
Some bias of this study needs to be pointed out. Because Stattic was injected into lateral cerebral ventricle, the inhibitory effect of Stattic was not specific to STAT3 in astrocytes. In future research, to obtain more convincing results, specific manipulation of STAT3 in astrocytes should be performed. Meanwhile, the role of astrocyte-secreted glial cell-derived neurotrophic factor (GDNF) should also be explained, as it is also very important in supporting neurogenesis. Other limitations include that we only used young mice instead of aged ones because aging is an unmodifiable risk factor for stroke and the aged mouse model is closer to human stroke [47]. We also acknowledge that young mice without comorbidities which are the major concern when treating real patients [48]. Further studies using aged or comorbid model should be performed.

Conclusion
Collectively, our findings indicate that ADPN treatment improves neurogenesis and neurological recovery after ischemic stroke. In addition, STAT3/BDNF signaling in astrocyte is required for ADPN-induced neural reparative effect. These results suggest that ADPN is a potential drug for the improvement of stroke recovery.